HTOutlier: Hardware Trojan detection with side-channel signature outlier identification

Abstract

Hardware Trojan (HT) is a growing concern for the semiconductor industry. As a non-invasive and inexpensive approach, side-channel analysis methods based on signatures such as power, current, or circuit delay are widely used for HT detection. However, the effectiveness of these methods is greatly challenged by the ever-increasing process variation (PV) effects with technology scaling. In this work, considering the inherent relationship among side-channel signatures in a chip, we formulate the HT detection problem as a signature outlier identification problem, and solve it by comparing each signature with an estimated value from other signatures. Experimental results on benchmark circuits show that the proposed technique is much more effective than existing solutions.